Re: [PATCH -V8 02/10] mm/numa: automatically generate node migration order

[Zi Yan <ziy@nvidia.com>]

[-- Attachment #1.1: Type: text/plain, Size: 10177 bytes --]

On 19 Jun 2021, at 4:18, Huang, Ying wrote:

> Zi Yan <ziy@nvidia.com> writes:
>
>> On 18 Jun 2021, at 2:15, Huang Ying wrote:
>>
>>> From: Dave Hansen <dave.hansen@linux.intel.com>
>>>
>>> When memory fills up on a node, memory contents can be
>>> automatically migrated to another node.  The biggest problems are
>>> knowing when to migrate and to where the migration should be
>>> targeted.
>>>
>>> The most straightforward way to generate the "to where" list would
>>> be to follow the page allocator fallback lists.  Those lists
>>> already tell us if memory is full where to look next.  It would
>>> also be logical to move memory in that order.
>>>
>>> But, the allocator fallback lists have a fatal flaw: most nodes
>>> appear in all the lists.  This would potentially lead to migration
>>> cycles (A->B, B->A, A->B, ...).
>>>
>>> Instead of using the allocator fallback lists directly, keep a
>>> separate node migration ordering.  But, reuse the same data used
>>> to generate page allocator fallback in the first place:
>>> find_next_best_node().
>>>
>>> This means that the firmware data used to populate node distances
>>> essentially dictates the ordering for now.  It should also be
>>> architecture-neutral since all NUMA architectures have a working
>>> find_next_best_node().
>>>
>>> The protocol for node_demotion[] access and writing is not
>>> standard.  It has no specific locking and is intended to be read
>>> locklessly.  Readers must take care to avoid observing changes
>>> that appear incoherent.  This was done so that node_demotion[]
>>> locking has no chance of becoming a bottleneck on large systems
>>> with lots of CPUs in direct reclaim.
>>>
>>> This code is unused for now.  It will be called later in the
>>> series.
>>>
>>> Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
>>> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
>>> Reviewed-by: Yang Shi <shy828301@gmail.com>
>>> Cc: Michal Hocko <mhocko@suse.com>
>>> Cc: Wei Xu <weixugc@google.com>
>>> Cc: David Rientjes <rientjes@google.com>
>>> Cc: Dan Williams <dan.j.williams@intel.com>
>>> Cc: David Hildenbrand <david@redhat.com>
>>> Cc: osalvador <osalvador@suse.de>
>>>
>>> --
>>>
>>> Changes from 20200122:
>>>  * Add big node_demotion[] comment
>>> Changes from 20210302:
>>>  * Fix typo in node_demotion[] comment
>>> ---
>>>  mm/internal.h   |   5 ++
>>>  mm/migrate.c    | 175 +++++++++++++++++++++++++++++++++++++++++++++++-
>>>  mm/page_alloc.c |   2 +-
>>>  3 files changed, 180 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/mm/internal.h b/mm/internal.h
>>> index 2f1182948aa6..0344cd78e170 100644
>>> --- a/mm/internal.h
>>> +++ b/mm/internal.h
>>> @@ -522,12 +522,17 @@ static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn,
>>>
>>>  #ifdef CONFIG_NUMA
>>>  extern int node_reclaim(struct pglist_data *, gfp_t, unsigned int);
>>> +extern int find_next_best_node(int node, nodemask_t *used_node_mask);
>>>  #else
>>>  static inline int node_reclaim(struct pglist_data *pgdat, gfp_t mask,
>>>  				unsigned int order)
>>>  {
>>>  	return NODE_RECLAIM_NOSCAN;
>>>  }
>>> +static inline int find_next_best_node(int node, nodemask_t *used_node_mask)
>>> +{
>>> +	return NUMA_NO_NODE;
>>> +}
>>>  #endif
>>>
>>>  extern int hwpoison_filter(struct page *p);
>>> diff --git a/mm/migrate.c b/mm/migrate.c
>>> index 6cab668132f9..111f8565f75d 100644
>>> --- a/mm/migrate.c
>>> +++ b/mm/migrate.c
>>> @@ -1136,6 +1136,44 @@ static int __unmap_and_move(struct page *page, struct page *newpage,
>>>  	return rc;
>>>  }
>>>
>>> +
>>> +/*
>>> + * node_demotion[] example:
>>> + *
>>> + * Consider a system with two sockets.  Each socket has
>>> + * three classes of memory attached: fast, medium and slow.
>>> + * Each memory class is placed in its own NUMA node.  The
>>> + * CPUs are placed in the node with the "fast" memory.  The
>>> + * 6 NUMA nodes (0-5) might be split among the sockets like
>>> + * this:
>>> + *
>>> + *	Socket A: 0, 1, 2
>>> + *	Socket B: 3, 4, 5
>>> + *
>>> + * When Node 0 fills up, its memory should be migrated to
>>> + * Node 1.  When Node 1 fills up, it should be migrated to
>>> + * Node 2.  The migration path start on the nodes with the
>>> + * processors (since allocations default to this node) and
>>> + * fast memory, progress through medium and end with the
>>> + * slow memory:
>>> + *
>>> + *	0 -> 1 -> 2 -> stop
>>> + *	3 -> 4 -> 5 -> stop
>>> + *
>>> + * This is represented in the node_demotion[] like this:
>>> + *
>>> + *	{  1, // Node 0 migrates to 1
>>> + *	   2, // Node 1 migrates to 2
>>> + *	  -1, // Node 2 does not migrate
>>> + *	   4, // Node 3 migrates to 4
>>> + *	   5, // Node 4 migrates to 5
>>> + *	  -1} // Node 5 does not migrate
>>> + */
>>> +
>>> +/*
>>> + * Writes to this array occur without locking.  READ_ONCE()
>>> + * is recommended for readers to ensure consistent reads.
>>> + */
>>>  static int node_demotion[MAX_NUMNODES] __read_mostly =
>>>  	{[0 ...  MAX_NUMNODES - 1] = NUMA_NO_NODE};
>>>
>>> @@ -1150,7 +1188,13 @@ static int node_demotion[MAX_NUMNODES] __read_mostly =
>>>   */
>>>  int next_demotion_node(int node)
>>>  {
>>> -	return node_demotion[node];
>>> +	/*
>>> +	 * node_demotion[] is updated without excluding
>>> +	 * this function from running.  READ_ONCE() avoids
>>> +	 * reading multiple, inconsistent 'node' values
>>> +	 * during an update.
>>> +	 */
>>> +	return READ_ONCE(node_demotion[node]);
>>>  }
>>
>> Is it necessary to have two separate patches to add node_demotion and
>> next_demotion_node() then modify it immediately? Maybe merge Patch 1 into 2?
>>
>> Hmm, I just checked Patch 3 and it changes node_demotion again and uses RCU.
>> I guess it might be much simpler to just introduce node_demotion with RCU
>> in this patch and Patch 3 only takes care of hotplug events.
>
> Hi, Dave,
>
> What do you think about this?
>
>>>
>>>  /*
>>> @@ -3144,3 +3188,132 @@ void migrate_vma_finalize(struct migrate_vma *migrate)
>>>  }
>>>  EXPORT_SYMBOL(migrate_vma_finalize);
>>>  #endif /* CONFIG_DEVICE_PRIVATE */
>>> +
>>> +/* Disable reclaim-based migration. */
>>> +static void disable_all_migrate_targets(void)
>>> +{
>>> +	int node;
>>> +
>>> +	for_each_online_node(node)
>>> +		node_demotion[node] = NUMA_NO_NODE;
>>> +}
>>> +
>>> +/*
>>> + * Find an automatic demotion target for 'node'.
>>> + * Failing here is OK.  It might just indicate
>>> + * being at the end of a chain.
>>> + */
>>> +static int establish_migrate_target(int node, nodemask_t *used)
>>> +{
>>> +	int migration_target;
>>> +
>>> +	/*
>>> +	 * Can not set a migration target on a
>>> +	 * node with it already set.
>>> +	 *
>>> +	 * No need for READ_ONCE() here since this
>>> +	 * in the write path for node_demotion[].
>>> +	 * This should be the only thread writing.
>>> +	 */
>>> +	if (node_demotion[node] != NUMA_NO_NODE)
>>> +		return NUMA_NO_NODE;
>>> +
>>> +	migration_target = find_next_best_node(node, used);
>>> +	if (migration_target == NUMA_NO_NODE)
>>> +		return NUMA_NO_NODE;
>>> +
>>> +	node_demotion[node] = migration_target;
>>> +
>>> +	return migration_target;
>>> +}
>>> +
>>> +/*
>>> + * When memory fills up on a node, memory contents can be
>>> + * automatically migrated to another node instead of
>>> + * discarded at reclaim.
>>> + *
>>> + * Establish a "migration path" which will start at nodes
>>> + * with CPUs and will follow the priorities used to build the
>>> + * page allocator zonelists.
>>> + *
>>> + * The difference here is that cycles must be avoided.  If
>>> + * node0 migrates to node1, then neither node1, nor anything
>>> + * node1 migrates to can migrate to node0.
>>> + *
>>> + * This function can run simultaneously with readers of
>>> + * node_demotion[].  However, it can not run simultaneously
>>> + * with itself.  Exclusion is provided by memory hotplug events
>>> + * being single-threaded.
>>> + */
>>> +static void __set_migration_target_nodes(void)
>>> +{
>>> +	nodemask_t next_pass	= NODE_MASK_NONE;
>>> +	nodemask_t this_pass	= NODE_MASK_NONE;
>>> +	nodemask_t used_targets = NODE_MASK_NONE;
>>> +	int node;
>>> +
>>> +	/*
>>> +	 * Avoid any oddities like cycles that could occur
>>> +	 * from changes in the topology.  This will leave
>>> +	 * a momentary gap when migration is disabled.
>>> +	 */
>>> +	disable_all_migrate_targets();
>>> +
>>> +	/*
>>> +	 * Ensure that the "disable" is visible across the system.
>>> +	 * Readers will see either a combination of before+disable
>>> +	 * state or disable+after.  They will never see before and
>>> +	 * after state together.
>>> +	 *
>>> +	 * The before+after state together might have cycles and
>>> +	 * could cause readers to do things like loop until this
>>> +	 * function finishes.  This ensures they can only see a
>>> +	 * single "bad" read and would, for instance, only loop
>>> +	 * once.
>>> +	 */
>>> +	smp_wmb();
>>> +
>>> +	/*
>>> +	 * Allocations go close to CPUs, first.  Assume that
>>> +	 * the migration path starts at the nodes with CPUs.
>>> +	 */
>>> +	next_pass = node_states[N_CPU];
>>
>> Is there a plan of allowing user to change where the migration
>> path starts? Or maybe one step further providing an interface
>> to allow user to specify the demotion path. Something like
>> /sys/devices/system/node/node*/node_demotion.
>
> I don't think that's necessary at least for now.  Do you know any real
> world use case for this?

In our P9+volta system, GPU memory is exposed as a NUMA node.
For the GPU workloads with data size greater than GPU memory size,
it will be very helpful to allow pages in GPU memory to be migrated/demoted
to CPU memory. With your current assumption, GPU memory -> CPU memory
demotion seems not possible, right? This should also apply to any
system with a device memory exposed as a NUMA node and workloads running
on the device and using CPU memory as a lower tier memory than the device
memory.


—
Best Regards,
Yan, Zi

[-- Attachment #1.2: Type: text/html, Size: 15393 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 854 bytes --]